hyPACK-2013 HPC GPU Cluster - Application Kernels
Image Processing - Edge Detection, Face Detecting, Image
Inpainting Methods
Image processing Kernels : Image Processing is gaining larger importance in a variety of application areas and it
deals with the manipulation and analysis of pictorial information. Image processing can
be a time consuming task and, parallel algorithms can be designed. Fast processing response
is a major requirement in many image processing applications. Even when the size of the
image is very large, typical vision systems involve real-time processing where a sequence
of image frames must be processed in a very short time.
-
Laplacian Edge Detection :
A common method for image processing
is pixel classification. Pixel classification defines a pixel's class based on one of its
features, in the case of edge detection (Laplacian Edge Detection); the feature examined is
its intensity versus the intensity of its neighbor pixels.
-
Face recognition:
The face recognition using machines is an active research topic and is widely used nowadays
in disciplines like image processing, pattern recognition, and computer vision. The main
interest is to acquire facial information from images and it provide clue to understand several
commercial systems such as security and surveillance. A face image analysis includes face
detection, recognition, tracking and rendering. As the basis for all other related image
analysis of human faces, face detection and tracking are of great importance. The major
challenges on the issues of facial recognition which identifies a relationship between two
basics variables of the process are reliability/accuracy of the technique being used and
computational cost of this technique. It is important to develop image analysis algorithms
that can meet real-time constraints along with data capturing devices and the vast amounts
of data that they generate.
-
Image inpainting :
Image inpainting refers to the process of reconstructing the original image which has been
damaged due to factors such as ageing, wear and tear and occlusion. The challenge lies in
the fact that the observer seeing the inpainted image should not be able to guess that the
image had been tampered with. Commonly used inpainting technique is an exemplar-based techniques
in which searching the best exemplar or the best patch in the undamaged portion of the image
that will be used for filling the damaged portions of the image.
PParallelisation : Image Processing Algorithms: The important steps in Laplace edge detection,
for parallelisation on GPUs will be discussed. The pixel classification in Edge dtection
defines a pixel's class based on one of its features. In the case of Laplacian Edge Detection,
the feature examined is its intensity versus the intensity
of its neighbor pixels. Also important steps in parallelisation of Face Detection and Image
inpainting on GPUs will be discussed.
Parallelisation: Fast Fourier Transformations (FFT) : An Overview of implementation of FFTs on HPC GPU Cluster
will be discussed.
HPC GPU Cluster :
In hyPACK-2013 workshop, a prototype HPC GPU cluster (CUDA /OpenCL enabled NVIDIA GPUs
& AMD-ATI OpenCL Prog. env) is used to solve
application kernels, that are based on Heterogeneous
programming model
In this workshop, programming and performance issues for applications on
HPC GPU Cluster will be discussed.
In laboratory session, a prototype Hybrid Heterogneous HPC GPU Cluster is made available,
which can address some of the heterogeneous computing workloads.
The HPC GPU Cluster can be made "adaptive" to the
application it is running, assigning the most effective resources in real-time as
per application demands, without requiring modifications to the application. One of the objectives of HPC GPU Cluster (hybrid computing system) is
to allocate resources of CPUs & GPUs in an optimal way to solve applications of different
characteristics.
|